Synchronous Averaging - The Trigger is Everything · 2019-10-30 · Synchronous Averaging? Reduce noise – Noise Reduction is proportional to the square root of the number of averages

Synchronous Averaging Synchronous Averaging -- The The Trigger is Everything...Trigger is Everything...

Gerry Priebe Gerry Priebe -- PrePre--BB--TekTekRepresenting Data Physics Corporation in North and Representing Data Physics Corporation in North and

South CarolinaSouth Carolina

Synchronous Averaging is used to Synchronous Averaging is used to detect a signal in uncorrelated noisedetect a signal in uncorrelated noise

Synchronous Averaging Synchronous Averaging –– Case StudyCase StudyCase Study – Paper Machine Calender Section

Used for changing and/or controlling sheet caliper (thickness)

Synchronous Averaging Synchronous Averaging –– Case Study Case Study Paper Machine Calender Paper Machine Calender –– Before Problem Before Problem

CorrectionCorrection

0°

180°

90°270°

1.0 in/s

-1.0 in/s

RMS: 0.0550989Max: 0.134449Min: -0.127059Pk to Pk: 0.261507Mean: 0.00723551

0°

180°

90°270°

1.0 in/s

-1.0 in/s


Vibration Synchronous to Top Roll – 1.64 ips, pk-pk

Vibration Synchronous to Bottom Roll – 0.26 ips, pk-pk

Synchronous Averaging Synchronous Averaging –– Case Study Case Study Paper Machine Calender Paper Machine Calender –– After Problem After Problem

CorrectionCorrection

0°

180°

90°270°

1.0 in/s

-1.0 in/s


0°

180°

90°270°

1.0 in/s

-1.0 in/s


Vibration Synchronous to Top Roll – 0.03 ips, pk-pk

Vibration Synchronous to Bottom Roll – 0.26 ips, pk-pk

Synchronous AveragingSynchronous Averaging

What is it?What is it?What are the benefits?What are the benefits?What can go wrong?What can go wrong?

Synchronous AveragingSynchronous Averaging

Synchronous averaging Synchronous averaging involves triggering data involves triggering data acquisition with respect to acquisition with respect to external events such as external events such as tachometer pulsestachometer pulses

Triggered acquisitions are Triggered acquisitions are averaged to remove the averaged to remove the contribution of contribution of asynchronous components asynchronous components thereby isolating the thereby isolating the contribution of a single contribution of a single synchronous componentsynchronous component

What is Averaging?What is Averaging?

A data frame of some arbitrary length (N) is gathered A data frame of some arbitrary length (N) is gathered into an array (Block 1)into an array (Block 1)

Frame 1


A second data frame of the same length (N) is gathered A second data frame of the same length (N) is gathered into a second array (Block 2)into a second array (Block 2)

Frame 1

+

Frame 2


These frames are then added together on a line by line These frames are then added together on a line by line basis to get the sum (Result)basis to get the sum (Result)

Frame 1

+

Frame 2

=

Result


If more averaging is desired each new data frame is If more averaging is desired each new data frame is added to the previous result to yield a New Resultadded to the previous result to yield a New Result

Result

+

New Frame

=

New Result


When the averaging is done the “right answer” is When the averaging is done the “right answer” is obtained by dividing by the number of averages (K)obtained by dividing by the number of averages (K)

Result

/

Number of Averages

=

Average


Eight records with a signal buried in noiseEight records with a signal buried in noise

S1 S2 S3 S4 S5 S6 S7 S8 S9 S10R1 -0.12323 -1.01524 1.381033 -1.92251 0.456146 0.392856 1.874923 -1.96605 -0.25703 -1.08908R2 0.369698 0.761429 0.920689 1.441884 -0.3041 1.598214 2.187615 -0.49151 0.514061 0.544538R3 -0.49293 0.25381 1.841378 0.961256 -0.45615 3.205358 2.125077 0.491513 -0.77109 -0.81681R4 -0.24647 -0.50762 -1.38103 0.480628 0.152049 2.803572 2.250154 1.966052 -1.02812 -0.54454R5 0.492931 -0.25381 -1.84138 -0.96126 -0.15205 3.607144 1.937462 0.983026 0.257031 0.272269R6 -0.3697 -0.76143 0.460344 -1.44188 0.608194 0.794642 1.749846 1.474539 0.771092 0.816807R7 0.123233 1.015239 -0.92069 1.922513 0.304097 2.401786 1.812385 -1.47454 -0.51406 -0.27227R8 0.246465 0.507619 -0.46034 -0.48063 -0.60819 1.196428 2.062538 -0.98303 1.028122 1.089076


Eight records with a signal buried in noiseEight records with a signal buried in noise

-3

-2

-1

0

1

2

3

4

1 2 3 4 5 6 7 8 9 10

R1R2R3R4R5R6R7R8


Eight records with a signal buried in noiseEight records with a signal buried in noiseTwo Averages:Two Averages:

A2

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

1 2 3 4 5 6 7 8 9 10

A2


Eight records with a signal buried in noiseEight records with a signal buried in noiseFourFour Averages:Averages: A4

-1

-0.5

0

0.5

1

1.5

2

2.5

1 2 3 4 5 6 7 8 9 10

A4


Eight records with a signal buried in noiseEight records with a signal buried in noiseEightEight Averages:Averages: A8

-0.5

0

0.5

1

1.5

2

2.5

1 2 3 4 5 6 7 8 9 10

A8

Some History?Some History?

The first Digital Signal Analyzer were The first Digital Signal Analyzer were Waveform Averagers (HP 5480A Circa Waveform Averagers (HP 5480A Circa 1967)1967)

External Trigger

Analog-to-Digital ConverterInput Signals

Digital Processing Circuits

Digital Data

Operator Controls

What are the benefits of What are the benefits of Synchronous Averaging?Synchronous Averaging?

Reduce noiseReduce noise–– Noise Reduction is Noise Reduction is

proportional to the proportional to the square root of the square root of the number of averages number of averages 11

Records Averaged

Noise Reduction

(dB)2 34 68 916 1232 1564 18

128 21256 24512 271024 302048 334096 368192 39

16384 4232768 45

1. “Synchronous Waveform Averaging: Magic Bullet for Noise”, Bob Masta, www.chipcenter.com

What are the benefits of What are the benefits of Synchronous Averaging?Synchronous Averaging?Isolates your analysis to only the rotor Isolates your analysis to only the rotor you are evaluatingyou are evaluating2 2 (maybe)(maybe)Will allow balancing in the presence of Will allow balancing in the presence of many nearby machinesmany nearby machines22

Allows analysis of speed varying Allows analysis of speed varying machines (maybe)machines (maybe)Time domain signals hidden in noise Time domain signals hidden in noise can be viewedcan be viewed

2. “Synchronous Time Averaging Applications”, Technical Associates of Charlotte, P.C., www.technicalassociates.net

When Synchronous Averaging When Synchronous Averaging won’t work?won’t work?

When the Signal being averaged does When the Signal being averaged does not appear at the same place not appear at the same place eacheach and and everyevery frameframe

When Synchronous Averaging When Synchronous Averaging won’t work?won’t work?

When the Signal being averaged does When the Signal being averaged does not appear at the same place not appear at the same place eacheach and and everyevery frameframe–– Trigger jitterTrigger jitter–– Signal jitterSignal jitter–– Poor trigger signal shapePoor trigger signal shape–– Noise on the trigger signalNoise on the trigger signal–– Synchronous noise Synchronous noise -- can’t be removedcan’t be removed

Trigger JitterTrigger Jitter

0 1.0 2.0 3.0 4.0 5.0 6.0-2.0

-1.5

-1.0

-500.0m

0

500.0m

1.0

1.5

2.0

sec

Rea

l, g

Trigger Signal - Single Record

0 1.0 2.0 3.0 4.0 5.0 6.0-1.0

-500.0m

0

500.0m

1.0

secR

eal,

g

Trigger Signal - 18 Averages

Detected by examining the average of the trigger signal

Examination of the first trigger pulse in the average shows that the pulse is not always in the exact same place. Therefore, the entire average is bad and it deteriorates the farther

you go into the data block

Signal JitterSignal Jitter

Trigger Signal - Single Record Trigger Signal - 18 Averages

Detected by examining the average of the trigger signal

Examination of the first trigger pulse in the average shows that the pulse is the correct shape. Therefore, the average appears to be starting at the same place each record.

However, the fact that the average of the trigger deteriorates as time increases suggests that even though the trigger is stable the time between pulses is not stable.

0 1.0 2.0 3.0 4.0 5.0 6.0-2.0

-1.5

-1.0

-500.0m

0

500.0m

1.0

1.5

2.0

sec

Rea

l, g

0 1.0 2.0 3.0 4.0 5.0 6.0-2.0

-1.5

-1.0

-500.0m

0

500.0m

1.0

1.5

2.0

secR

eal,

g

What Causes Jitter?What Causes Jitter?

Relative motion between the machine Relative motion between the machine and the photoand the photo--tachometertachometer22

Laser Tachometer

Poor Trigger Signal ShapePoor Trigger Signal Shape

-1.0 0 1.0 2.0 3.0 4.0 5.0-800.0m

-600.0m

-400.0m

-200.0m

0

200.0m

400.0m

600.0m

800.0m

sec

Rea

l, V

Noise on the Trigger SignalNoise on the Trigger Signal

0 1.0 2.0 3.0 4.0 5.0 6.0-400.0m

-300.0m

-200.0m

-100.0m

0

100.0m

200.0m

300.0m

400.0m

sec

Rea

l, V

Single Capture

Synchronous noiseSynchronous noise

0 1.0 2.0 3.0 4.0 5.0 6.0-4.0

-3.0

-2.0

-1.0

0

1.0

2.0

3.0

4.0

sec

Rea

l, g

10+ Averages

RMS: 0.3960780 1.0 2.0 3.0 4.0 5.0 6.0

-4.0

-3.0

-2.0

-1.0

0

1.0

2.0

3.0

4.0

sec

Rea

l, g

One Average

RMS: 0.400377

Identified by the absence of a Signal to Noise Ratio (SNR) improvement


When the signal being averaged does When the signal being averaged does not appear at the same place not appear at the same place eacheach and and everyevery frameframe–– Poor trigger processing in the analyzerPoor trigger processing in the analyzer


Poor trigger processing in analyzerPoor trigger processing in analyzerDP Mobilyzer DSA

DPMentorTraining

Aid

Signal

Trigger

Square waveformof frequency F1/4

Square waveformof frequency F1


Test One: Trigger on Input Channel Test One: Trigger on Input Channel DP Mobilyzer DSA

DPMentorTraining

Aid

Signal

Trigger




Test One: ResultsTest One: Results

750.0m700.0m 750.0m-8.0

-6.0

-4.0

-2.0

0

2.0

4.0

6.0

8.0

sec

Rea

l, V

Signal - Channel Two Trigger

Ch 1

Ch 2

Signal

Trigger



Display of 10+ records from T0 + 0.7 seconds superimposed in a single display


Test Two: External Trigger InputTest Two: External Trigger InputDP Mobilyzer DSA

DPMentorTraining

Aid

Signal

Trigger




Test Two: ResultsTest Two: Results

750.0m700.0m 750.0m-8.0

-6.0

-4.0

-2.0

0

2.0

4.0

6.0

8.0

sec

Rea

l, V

Signal - External Trigger

Ch 1

ExtTrigger

Signal

Trigger



Display of 10+ records from T0 + 0.7 seconds superimposed in a single display


–– What is different between triggering on an What is different between triggering on an input channelinput channel and the and the external trigger external trigger inputinput??


The Sampling FrequencyThe Sampling Frequency–– When the trigger is connected to the When the trigger is connected to the input input

channelchannel, it is sampled at the same rate as , it is sampled at the same rate as the the signal channelsignal channel

In the case of our example the Sampling In the case of our example the Sampling Frequency FFrequency Fss = ~ 2,600 Hz= ~ 2,600 HzTherefore, Delta T = 400,000 Therefore, Delta T = 400,000 nSnSThe trigger jitter = The trigger jitter = ++ 200,000 200,000 nSnS


The Sampling FrequencyThe Sampling Frequency–– When the trigger is connected to the When the trigger is connected to the

external trigger inputexternal trigger input it is sampled at a it is sampled at a much higher rate to minimize the jittermuch higher rate to minimize the jitter

In the case of the DP In the case of the DP MobilyzerMobilyzer the external the external trigger is sampled at a frequency of trigger is sampled at a frequency of 10 MHz10 MHzTherefore, Delta T = 100 Therefore, Delta T = 100 nSecnSecThe trigger jitter = The trigger jitter = ++ 50 50 nSecnSec

Frequency Domain ErrorsFrequency Domain Errors

Consider the analysis of a Square WaveformConsider the analysis of a Square Waveform–– If a perfect square wave is analyzed:If a perfect square wave is analyzed:

Fundamental (HFundamental (H11) = 1) = 1Third Harmonic (HThird Harmonic (H33) = 1/3) = 1/3Fifth Harmonic (HFifth Harmonic (H55) = 1/5) = 1/5Seventh Harmonic (HSeventh Harmonic (H77) = 1/7) = 1/7Ninth Harmonic (HNinth Harmonic (H99) = 1/9) = 1/9......

Frequency Domain ErrorsFrequency Domain Errors

0.1

1

10

Harmonic

The Odd Harmonics of a Square Waveform

Theory (T)

Average with ADC trigger (AC)

Average with External Trigger (AE)

One record with ADC trigger (RC)

One record with External Trigger (RE)

Note that even though AE, RC & RE don’t agree with theory (T) they do agree with each other

How big is the error?How big is the error?% Error of Each Harmonic

-0.33%-2.57%

-7.18%

-13.87%

-21.61%

-30.69%

-40.00%

-48.82%

-56.84%

-80.00%

-70.00%

-60.00%

-50.00%

-40.00%

-30.00%

-20.00%

-10.00%

0.00%

1 3 5 7 9 11 13 15 17 19

Error

How do I know?How do I know?

Look for evidence of jitterLook for evidence of jitter–– Examine the average of the tachometerExamine the average of the tachometer

Is the average of the tachometer good at the Is the average of the tachometer good at the beginning and bad at the end?beginning and bad at the end?

–– Order tracking may be neededOrder tracking may be needed

Is the shape as expected?Is the shape as expected?–– Look at the leading and trailing edge for signs of poor Look at the leading and trailing edge for signs of poor

averagingaveraging–– Examine the value of any overExamine the value of any over-- or underor under--shootshoot

Order TrackingOrder Tracking vsvs RPMRPM--relatedrelated

RPMRPM--related for slowly changing machine speed, constant related for slowly changing machine speed, constant speed within a revolutionspeed within a revolution

–– Needs only a once per revNeeds only a once per rev tachtach

Order tracking for quickly changing machine speed, or Order tracking for quickly changing machine speed, or speed that varies though the cyclespeed that varies though the cycle

–– Need several pulses per revolutionNeed several pulses per revolution

Order TrackingOrder Tracking

How is Triggering done?How is Triggering done?

Check the analyzer specificationsCheck the analyzer specifications–– How does the trigger input circuit work?How does the trigger input circuit work?–– Determine the speed of the trigger sample clockDetermine the speed of the trigger sample clock

Abacus –– Data Flow

DSP32

8

8

NET

recording

IN

TRIG

OUT

INPUTS measure voltage

TRIGGERS use voltage to measure time



Check the analyzer specificationsCheck the analyzer specifications–– Is the Is the trigger thresholdtrigger threshold adjustable?adjustable?–– Is the Is the SlopeSlope (+ or (+ or --) selectable?) selectable?–– Is trigger Is trigger HysteresisHysteresis available and adjustable?available and adjustable?–– Is trigger Is trigger Hold OffHold Off available and adjustable?available and adjustable?

Typically these features are not available for Typically these features are not available for analyzers with analyzers with poor triggeringpoor triggering


DP Abacus DP Abacus ((MobilyzerMobilyzer II) = 25 II) = 25 MHz (i.e. MHz (i.e. ++ 2020 nSecnSec) )


AgilentAgilent E1432A = 20 MHz E1432A = 20 MHz (i.e. (i.e. ++ 2525 nSecnSec))


DPDP MobilyzerMobilyzer = 10 MHz (i.e. = 10 MHz (i.e. ++ 5050 nSecnSec))


DP ACE = 51.2 kHz (i.e. DP ACE = 51.2 kHz (i.e. ++ 9,7009,700 nSecnSec))


DP Abacus (DP Abacus (MobilyzerMobilyzer II) = 25 MHz (i.e. II) = 25 MHz (i.e. ++ 2020 nSecnSec))AgilentAgilent E1432A = 20 MHz (i.e. E1432A = 20 MHz (i.e. ++ 2525 nSecnSec))DP DP Mobilyzer Mobilyzer = 10 MHz (i.e. = 10 MHz (i.e. ++ 50 50 nSecnSec))DP ACE = 51.2 kHz (i.e. DP ACE = 51.2 kHz (i.e. ++ 9,700 9,700 nSecnSec))DAT Tape (5 kHz) = 12.8 kHz (i.e. DAT Tape (5 kHz) = 12.8 kHz (i.e. ++ 39,000 39,000 nSecnSec))Input Channel Trigger = 2.6 kHz Input Channel Trigger = 2.6 kHz (i.e. (i.e. ++ 200,000 200,000 nSecnSec))YOUR ANALYZER = ?YOUR ANALYZER = ?

ConclusionConclusion

JitterJitter effects can be effects can be subtle & subtle & hiddenhiddenhiddenNot all analyzers are created Not all analyzers are created eQuaLeQuaLThe same problems arise for digital The same problems arise for digital tape recorderstape recordersBest analysis is “live” and on the scene Best analysis is “live” and on the scene with a with a well designed DSAwell designed DSA

End of PresentationEnd of Presentation

Synchronous Averaging - The Trigger is Everything · 2019-10-30 · Synchronous Averaging? Reduce noise – Noise Reduction is proportional to the square root of the number of averages

Documents